framework,version,device,op_name,kernel_source,bmm_dtype,num_tokens,num_heads,latency
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,128,0.010899200290441512
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,64,0.008736000210046769
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,32,0.008828800171613693
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,128,0.006771200150251388
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,32,0.004707200080156326
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,16,0.0046016000211238865
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,8,0.0045311998575925825
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,8,0.007686399668455124
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,4,0.008675199747085572
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,2,0.004662400111556053
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,2,0.008630400151014328
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,1,0.004499199986457825
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,1,0.00868159979581833
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,128,0.006812799721956253
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,4,0.004608000069856644
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,64,0.005401600152254105
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1,64,0.006163199990987777
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,32,0.00480320006608963
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,128,0.01071999967098236
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,16,0.004800000041723251
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,16,0.008761599659919739
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,8,0.0045471999794244765
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,8,0.008617600053548813
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,4,0.0055808000266551975
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,4,0.008716800063848496
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,2,0.004652800038456917
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,2,0.008790399879217148
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2,1,0.004614400118589402
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,1,0.00859839990735054
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,128,0.006864000111818314
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,128,0.010732799768447876
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,64,0.0046271998435258865
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,64,0.008591999858617782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,64,0.008748800307512284
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,32,0.008607999980449676
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,16,0.0051968000829219815
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,16,0.008710400015115739
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,8,0.004572800174355507
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,8,0.008687999844551087
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,4,0.004860800132155419
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,4,0.008668799698352814
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,2,0.004687999933958053
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1,16,0.008377599716186523
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,2,0.008720000088214875
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,1,0.004726399853825569
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4,1,0.008752000331878663
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,128,0.006665600091218948
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,128,0.01069760024547577
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,64,0.006217600032687187
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,64,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,32,0.004783999919891357
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,32,0.008662399649620057
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,16,0.004790399968624115
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,16,0.008700799942016602
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,8,0.008607999980449676
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,4,0.0046431999653577805
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,4,0.008816000074148178
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,2,0.004623999819159508
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,2,0.008697599917650223
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,1,0.004652800038456917
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8,1,0.00868159979581833
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,128,0.006668800115585327
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4,32,0.004560000076889992
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,64,0.00605119988322258
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,64,0.008851200342178345
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,32,0.004575999826192856
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,32,0.008585599809885025
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,16,0.004656000062823296
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,16,0.008614400029182434
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,8,0.004560000076889992
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,8,0.008486399799585343
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,4,0.004560000076889992
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,4,0.0086496002972126
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,2,0.004678399860858917
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8,8,0.0049056001007556915
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,16,1,0.004761600121855736
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,1,0.008329600095748901
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,128,0.008342400193214417
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,128,0.014886400103569031
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,64,0.006224000081419945
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,64,0.010784000158309937
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2,32,0.008614400029182434
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,32,0.005052800104022026
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,32,0.008963199704885483
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,16,0.008876799792051315
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,8,0.0047807998955249785
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,8,0.00854720026254654
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,4,0.004800000041723251
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,4,0.00840959995985031
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,2,0.004665600135922432
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,2,0.00859839990735054
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,1,0.005177599936723709
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,32,1,0.008377599716186523
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,128,0.008505599945783615
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,128,0.012745599448680877
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,128,0.01887679994106293
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,64,0.00655680000782013
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,64,0.012736000120639801
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,32,0.004774399846792221
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,16,0.004745600000023842
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,16,2,0.008671999722719193
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,16,0.008630400151014328
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,8,0.004691199958324432
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,8,0.008563199639320373
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,4,0.004579199850559235
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,4,0.00852160006761551
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,2,0.004515200108289719
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,2,0.00857279971241951
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,48,1,0.00456320010125637
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,32,16,0.00469760000705719
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,1,0.008575999736785888
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,128,0.006825599819421768
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,128,0.020985600352287293
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,64,0.006393600255250931
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,32,0.006172800064086914
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,32,0.010742399841547012
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,16,0.004774399846792221
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,16,0.008873599767684936
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,8,0.004783999919891357
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,8,0.008828800171613693
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,4,0.00469760000705719
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,4,0.008313599973917007
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,2,0.004732799902558327
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,2,0.008460800349712371
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,48,32,0.010556799918413162
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,64,1,0.004745600000023842
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,1,0.008416000008583068
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,128,0.008339200168848038
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,64,0.0063680000603199005
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,128,0.022915199398994446
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,32,0.006070400029420853
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,64,0.014988799393177033
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,32,0.010911999642848969
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,16,0.010416000336408614
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,8,0.004822399839758873
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,8,0.008739200234413148
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,4,0.004675199836492538
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,4,0.008460800349712371
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,2,0.004812800139188766
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,2,0.008716800063848496
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,1,0.004870399832725525
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,80,1,0.008591999858617782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,128,0.008531200140714646
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,128,0.025171199440956117
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,64,0.006595200300216675
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,64,0.01704320013523102
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,64,64,0.014745600521564484
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,32,0.012748800218105316
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,16,0.00453759990632534
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,16,0.010623999685049058
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,8,0.006486400216817856
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,8,0.008607999980449676
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,4,0.004467200115323066
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,4,0.008627200126647949
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,2,0.006716799736022949
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,2,0.008508799970149994
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,96,1,0.008620800077915191
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,128,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,128,0.029232001304626463
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,64,0.0065311998128890995
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,64,0.018848000466823576
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,32,0.006598400324583054
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,32,0.013833600282669067
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,16,0.004732799902558327
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,16,0.010630399733781815
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,8,0.004543999955058098
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,8,0.008553600311279297
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,4,0.004569600149989128
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,4,0.008550400286912918
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,2,0.004553600028157234
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,2,0.008537600189447403
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,128,1,0.004656000062823296
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,128,1,0.008752000331878663
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,128,0.00878399983048439
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,128,0.03538880050182343
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,64,0.006579200178384781
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,32,0.006140799820423126
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,32,0.006329599767923355
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,32,0.014972800016403198
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,16,0.006415999680757523
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,16,0.010684800148010255
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,8,0.004716800153255462
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,8,0.009827200323343277
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,96,1,0.004572800174355507
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,4,0.004694399982690811
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,2,0.0047807998955249785
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,2,0.008611200004816055
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,160,1,0.005401600152254105
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,1,0.008342400193214417
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,128,0.010678400099277497
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,128,0.04142079949378967
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,80,16,0.0048351999372243885
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,64,0.008268799632787704
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,32,0.006332799792289734
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,64,0.02497279942035675
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,32,0.016758400201797485
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,16,0.01276479959487915
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,8,0.005744000151753426
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,8,0.010678400099277497
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,4,0.004665600135922432
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,64,0.0209184005856514
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,4,0.008540800213813782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,2,0.004675199836492538
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,2,0.008582399785518646
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,1,0.004527999833226204
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,192,1,0.008540800213813782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,128,0.011283200234174728
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,128,0.05162559747695923
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,160,4,0.009286399930715561
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,64,0.00862400010228157
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,32,0.006592000275850296
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,64,0.03156799972057343
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,32,0.018806399405002595
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,16,0.006771200150251388
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,16,0.014416000247001648
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,8,0.006454399973154068
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,8,0.012080000340938568
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,4,0.005667199939489364
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,4,0.008591999858617782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,2,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,1,0.00591679997742176
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,256,1,0.008486399799585343
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,192,16,0.006233600154519081
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,128,0.012748800218105316
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,64,0.008591999858617782
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,128,0.06402559876441956
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,64,0.03535040020942688
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,32,0.006963200122117996
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,16,0.006358399987220764
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,32,0.021049599349498748
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,8,0.0057599999010562895
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,8,0.010812799632549285
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,4,0.004556800052523613
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,4,0.008787199854850769
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,2,0.00448639988899231
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,2,0.008697599917650223
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,320,1,0.00451200008392334
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,1,0.008566399663686752
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,128,0.01483519971370697
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,128,0.07208639979362488
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,64,0.011107199639081956
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,64,0.03989120125770569
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,32,0.0075552001595497135
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,32,0.025084799528121947
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,16,0.006540799885988236
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,16,0.016969600319862367
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,8,0.006473600119352341
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,8,0.012630400061607362
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,4,0.010611200332641601
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,256,2,0.005455999821424484
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,2,0.008662399649620057
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,1,0.004595199972391129
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,384,1,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,128,0.01883520036935806
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,128,0.09452160000801087
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,320,16,0.014815999567508698
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,64,0.012604799866676331
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,32,0.0086496002972126
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,64,0.05191680192947388
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,16,0.00663359984755516
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,32,0.029046401381492615
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,8,0.006601600348949433
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,16,0.01903360038995743
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,8,0.013148799538612366
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,4,0.01064639985561371
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,2,0.004681599885225296
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,2,0.00896959975361824
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,1,0.00456320010125637
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,512,1,0.008828800171613693
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,128,0.027055999636650084
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,128,0.13766720294952392
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,64,0.014575999975204468
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,64,0.07228800058364868
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,32,0.01085119992494583
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,32,0.041731199622154234
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,4,0.0060479998588562015
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,16,0.0251008003950119
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,384,2,0.004579199850559235
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,8,0.006694400310516357
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,8,0.016883200407028197
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,4,0.012716799974441528
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,2,0.0046847999095916745
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,2,0.01066880002617836
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,1,0.0045855998992919925
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,768,1,0.008742400258779526
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,128,0.03487359881401062
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,128,0.1826815962791443
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,64,0.016764800250530242
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,64,0.0946399986743927
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,32,0.0106175996363163
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,32,0.05156800150871277
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,16,0.008694399893283845
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,16,0.030582401156425475
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,8,0.006579200178384781
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,8,0.01887679994106293
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,4,0.006393600255250931
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,4,0.013260799646377563
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,2,0.004604800045490265
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,2,0.010627199709415436
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1024,1,0.004479999840259552
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,16,0.008620800077915191
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1024,1,0.008643200248479843
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,128,0.045440000295639035
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,768,4,0.006332799792289734
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,128,0.27087039947509767
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,64,0.02630079984664917
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,32,0.014595200121402741
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,32,0.0741823971271515
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,16,0.010844799876213073
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,16,0.041465601325035094
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,8,0.008294399827718735
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,8,0.024979199469089507
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,4,0.006611199676990509
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,4,0.01687680035829544
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,2,0.0064640000462532045
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,512,4,0.006067200005054474
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,1536,1,0.0060095999389886854
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,1,0.010351999849081039
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,128,0.06010559797286987
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,128,0.3656512022018433
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,64,0.03319680094718933
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,64,0.1867840051651001
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,32,0.016873599588871004
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,32,0.09680320024490356
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,16,0.010953599959611893
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,16,0.051823997497558595
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,8,0.008796799927949905
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,8,0.03118079900741577
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,4,0.006771200150251388
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,4,0.018934400379657747
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,64,0.13782720565795897
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,2,0.0062720000743865965
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,2048,1,0.005395200103521347
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,1,0.010902400314807891
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,128,0.08438400030136109
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,128,0.5455167770385743
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,64,0.04493759870529175
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,1536,2,0.012831999361515046
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,64,0.2770751953125
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,32,0.13967039585113525
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,16,0.07401599884033203
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,8,0.010716799646615982
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,8,0.04166400134563446
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,4,0.008505599945783615
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,4,0.025174400210380553
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,2,0.006691200286149978
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,2,0.016681599617004394
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,1,0.006326399743556976
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,3072,1,0.012992000579833985
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,128,0.11095679998397827
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,128,0.7191232204437256
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,64,0.058246397972106935
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,2048,2,0.012956799566745758
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,32,0.031939199566841124
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,32,0.18068480491638184
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,16,0.01679359972476959
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,16,0.09474239945411682
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,8,0.010755199939012527
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,8,0.051910400390625
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,4,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,16,0.014428800344467163
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,4,0.029212799668312073
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,2,0.006780800223350525
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,4096,1,0.006358399987220764
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,2,0.018905599415302277
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,1,0.014467200636863709
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,128,0.15773439407348633
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,64,0.08146880269050598
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,64,0.5318431854248047
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,128,1.0964256286621095
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,32,0.27631680965423583
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,16,0.025600001215934753
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,16,0.13971199989318847
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,8,0.014579200744628906
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,8,0.07422720193862915
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,4,0.01080320030450821
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,4,0.043375998735427856
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,2,0.008489599823951722
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,4096,64,0.3622368097305298
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,2,0.02513279914855957
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,1,0.006595200300216675
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,6144,1,0.01682240068912506
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,128,0.20808639526367187
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,64,0.10714240074157715
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,64,0.7170815944671631
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,32,0.05764480233192444
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,3072,32,0.026256000995635985
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,16,0.03315199911594391
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,32,0.36175999641418455
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,8,0.017020800709724428
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,16,0.18223680257797242
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,4,0.01117440015077591
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,8,0.09674559831619263
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,4,0.051999998092651364
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,2,0.008675199747085572
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,2,0.0293503999710083
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,1,0.01915840059518814
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,6144,32,0.043558400869369504
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,fp8,8192,128,1.4817503929138183
SGLang,0.5.9,NVIDIA GB300,mla_gen_pre,default,float16,8192,1,0.006707199662923813
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,128,0.006588800251483918
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,64,0.0061471998691558834
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,128,0.008787199854850769
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,32,0.00846719965338707
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,64,0.0084927998483181
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,16,0.008367999643087386
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,8,0.006121600046753883
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,8,0.008867199718952178
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,4,0.006137600168585777
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,4,0.008463999629020691
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,2,0.00631679967045784
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,2,0.008515200018882752
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,1,0.0062015999108552934
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1,1,0.008416000008583068
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,128,0.0064640000462532045
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,32,0.006335999816656113
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,64,0.004687999933958053
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,128,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,32,0.006431999802589417
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1,16,0.006265600025653839
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,16,0.006217600032687187
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,16,0.008646400272846222
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,8,0.006015999987721443
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,8,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,4,0.006099199876189232
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,4,0.008617600053548813
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,2,0.005862399935722351
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,2,0.008262400329113007
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2,1,0.005980800092220307
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,1,0.008326400071382523
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,128,0.006255999952554703
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,128,0.008694399893283845
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,64,0.006092799827456474
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,64,0.008745600283145905
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,32,0.006207999959588051
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,32,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,16,0.006083200126886368
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,16,0.00859839990735054
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,8,0.00626240000128746
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,8,0.008316799998283386
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,4,0.004790399968624115
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,4,0.008607999980449676
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,2,0.004931199923157692
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4,1,0.004742399975657463
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,1,0.008323200047016144
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,128,0.006358399987220764
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,128,0.010489600151777268
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,64,0.0061471998691558834
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,64,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,32,0.006294400244951248
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,32,0.008656000345945358
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,16,0.006249599903821945
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,16,0.008703999966382981
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,8,0.006137600168585777
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,8,0.008265600353479386
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,4,0.0047680001705884935
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,4,0.008560000360012055
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,2,0.004662400111556053
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,2,0.00841279998421669
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8,1,0.004598399996757508
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8,1,0.008323200047016144
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,128,0.006371200084686279
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,128,0.011139199882745743
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,64,0.0064191997051239015
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,64,0.00880960002541542
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,32,0.005446400120854377
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,32,0.008832000195980072
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,16,0.004735999926924706
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,16,0.008444800227880477
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,8,0.006387200206518173
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,32,0.008495999872684479
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,8,0.008748800307512284
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,4,0.006204799935221672
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,2,0.006150399893522262
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,2,0.00838399976491928
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16,1,0.005836800113320351
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,1,0.008428800106048583
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,128,0.006707199662923813
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,128,0.01449279934167862
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,64,0.006326399743556976
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,64,0.010927999764680863
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,32,0.006380800157785416
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,32,0.008899199962615966
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,16,0.005475199967622757
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,16,0.008806400001049042
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,8,0.00613120011985302
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,8,0.008873599767684936
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,4,0.006207999959588051
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,4,0.008419200032949447
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,2,0.005635200068354607
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,2,0.00867839977145195
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,32,1,0.005392000079154968
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,32,1,0.008399999886751174
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,128,0.006822399795055389
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,128,0.01687680035829544
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,64,0.006348799914121628
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,64,0.012771199643611907
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,32,0.006319999694824219
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,32,0.010515200346708298
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2,64,0.008691199868917466
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,16,0.00589120015501976
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,16,0.008937600255012512
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,8,0.006265600025653839
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,8,0.008966399729251862
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,4,0.006297600269317627
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,4,0.008499199897050858
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,2,0.004924799874424934
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,2,0.008457600325345992
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,48,1,0.004822399839758873
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16,4,0.008566399663686752
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,128,0.006675200164318084
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,128,0.018960000574588777
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,64,0.006249599903821945
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,64,0.014473600685596466
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,32,0.006032000109553337
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,32,0.010691200196743012
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,16,0.0062431998550891874
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,16,0.008873599767684936
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,8,0.004975999891757965
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,8,0.008588799834251403
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,4,0.004745600000023842
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,4,0.008371199667453765
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,2,0.004700800031423568
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,2,0.008483199775218964
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,64,1,0.005923200026154518
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,64,1,0.008483199775218964
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,128,0.006803199648857117
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,128,0.021062399446964263
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,64,0.006508799642324448
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,64,0.014854399859905243
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,32,0.00642239972949028
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,32,0.010822399705648422
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,16,0.006297600269317627
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,16,0.010441599786281586
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,8,0.0062304001301527025
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,8,0.008851200342178345
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,4,0.006111999973654747
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,4,0.008729600161314011
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,2,0.0062015999108552934
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,48,1,0.008284799754619598
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4,2,0.007747200131416321
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,80,1,0.00613120011985302
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,128,0.006844799965620041
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,1,0.008614400029182434
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,128,0.023100799322128295
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,64,0.016748799383640288
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,32,0.00639680027961731
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,32,0.01281919926404953
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,16,0.006313599646091461
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,16,0.010416000336408614
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,8,0.006198399886488915
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,8,0.00886079967021942
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,4,0.005542400106787681
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,4,0.00862400010228157
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,2,0.0049056001007556915
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,2,0.008451200276613235
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,1,0.004940799996256828
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,96,1,0.008540800213813782
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,128,0.0074432000517845156
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,128,0.028966400027275085
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,64,0.006259199976921081
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,64,0.01881600022315979
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,32,0.006278400123119354
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,32,0.014691199362277984
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,16,0.0062720000743865965
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,16,0.0108255997300148
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,8,0.009078399837017059
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,8,0.008950400352478027
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,4,0.005215999856591225
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,4,0.008848000317811966
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,2,0.005471999943256378
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,2,0.008345600217580795
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,128,1,0.005584000051021576
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,80,2,0.008336000144481659
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,128,1,0.008502399921417237
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,128,0.008399999886751174
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,128,0.03329600095748901
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,64,0.006377600133419037
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,64,0.02099200040102005
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,32,0.006304000318050384
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,16,0.006335999816656113
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,32,0.015007999539375306
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,8,0.006070400029420853
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,16,0.010796800255775452
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,8,0.01051200032234192
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,4,0.004950400069355964
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,4,0.008854400366544724
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,2,0.004944000020623207
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,2,0.008825600147247314
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,160,1,0.008742400258779526
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,128,0.008595199882984161
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,128,0.03931199908256531
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,64,0.00663359984755516
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,64,0.02313600033521652
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,32,0.006284800171852112
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,32,0.016198399662971496
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,16,0.006310400366783142
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,16,0.012627199292182922
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,8,0.006204799935221672
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,8,0.010678400099277497
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,96,64,0.006403200328350067
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,4,0.006940799951553345
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,4,0.00859839990735054
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,2,0.007311999797821045
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,2,0.008585599809885025
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,192,1,0.007062400132417679
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,192,1,0.00867839977145195
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,128,0.009151999652385712
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,128,0.049481600522994995
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,64,0.0072063997387886046
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,64,0.028992000222206115
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,32,0.006595200300216675
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,32,0.01879040002822876
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,16,0.006809599697589874
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,16,0.01300159990787506
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,8,0.006428799778223038
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,8,0.010684800148010255
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,4,0.006451199948787689
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,4,0.008742400258779526
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,2,0.005536000058054924
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,2,0.00857279971241951
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,256,1,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,128,0.009014400094747544
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,128,0.05977920293807983
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,64,0.0065151996910572055
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,64,0.03319360017776489
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,32,0.006483200192451477
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,160,1,0.004665600135922432
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,32,0.020819200575351714
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,16,0.006140799820423126
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,16,0.014681600034236908
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,8,0.006224000081419945
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,4,0.006255999952554703
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,8,0.010806400328874588
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,4,0.010620799660682679
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,2,0.0045471999794244765
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,2,0.00897279977798462
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,320,1,0.004582399874925614
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,320,1,0.008841600269079208
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,128,0.010716799646615982
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,64,0.008556800335645676
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,128,0.07200639843940734
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,64,0.03930880129337311
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,32,0.02292799949645996
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,16,0.006387200206518173
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,16,0.016649599373340606
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,8,0.0062015999108552934
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,8,0.012748800218105316
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,4,0.004704000055789947
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,4,0.010652799904346467
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,2,0.004502400010824204
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,2,0.008607999980449676
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,1,0.004508800059556961
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,384,1,0.008550400286912918
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,128,0.014950400590896607
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,128,0.09463359713554383
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,64,0.008601599931716919
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,256,1,0.004732799902558327
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,64,0.0496288001537323
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,32,0.006694400310516357
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,16,0.00647680014371872
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,16,0.018803200125694274
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,8,0.006403200328350067
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,8,0.01286720037460327
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,4,0.004540799930691719
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,4,0.010585600137710571
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,2,0.005878400057554245
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,512,1,0.005129599943757057
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,2,0.00873280018568039
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,1,0.008963199704885483
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,128,0.02688960134983063
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,128,0.1398911952972412
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,64,0.010716799646615982
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,32,0.008595199882984161
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,64,0.07107840180397033
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,32,0.03934719860553741
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,384,32,0.006787200272083282
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,16,0.0229312002658844
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,8,0.006150399893522262
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,8,0.01594880074262619
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,4,0.00628800019621849
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,2,0.008803199976682663
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,4,0.012608000636100769
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,2,0.010367999970912933
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,1,0.004569600149989128
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,768,1,0.008796799927949905
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,128,0.03320960104465485
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,64,0.012745599448680877
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,128,0.1836832046508789
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,32,0.0088128000497818
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,64,0.09431999921798706
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,16,0.006518399715423584
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,32,0.049481600522994995
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,16,0.028972798585891725
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,8,0.006348799914121628
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,8,0.018886399269104005
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,4,0.006345599889755249
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,2,0.006268800050020218
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,2,0.01090560033917427
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1024,1,0.006486400216817856
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,512,32,0.02900800108909607
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,1,0.008697599917650223
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,128,0.04671359956264496
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,128,0.27225921154022215
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,64,0.025536000728607178
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,64,0.13982399702072143
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,32,0.010675200074911118
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,32,0.07014399766921997
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,16,0.008604799956083297
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,768,16,0.0067552000284194945
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,8,0.006636799871921539
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,8,0.023363199830055238
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,4,0.006188800185918808
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,4,0.016521599888801575
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,2,0.00658240020275116
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,2,0.01286720037460327
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,1536,1,0.006329599767923355
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,1,0.010361599922180175
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,128,0.06203200221061707
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,128,0.36193599700927737
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1024,4,0.012966400384902954
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,64,0.03365119993686676
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,32,0.013052800297737121
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,64,0.1827839970588684
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,16,0.008771199733018875
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,16,0.04967359900474548
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,8,0.006764800101518631
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,8,0.02908799946308136
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,4,0.006505600363016129
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,4,0.018822400271892546
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,2,0.0066431999206542965
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,2,0.012800000607967377
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,2048,1,0.004566400125622749
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,1,0.01066880002617836
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,128,0.08314239978790283
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,128,0.5495488166809082
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,1536,16,0.03930560052394867
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,64,0.2710655927658081
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,32,0.024905599653720856
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,32,0.13888319730758666
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,16,0.07009919881820678
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,16,0.010953599959611893
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,8,0.008416000008583068
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,4,0.006588800251483918
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,4,0.023014399409294128
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,2,0.0062720000743865965
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,2,0.016764800250530242
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,1,0.006691200286149978
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,1,0.012643200159072877
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,128,0.11492799520492554
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,2048,32,0.09470400214195251
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,64,0.062294399738311766
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,128,0.7456384181976319
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,32,0.033155199885368344
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,64,0.3630239963531494
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,16,0.012768000364303589
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,32,0.1833343982696533
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,16,0.09485120177268982
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,8,0.008755200356245042
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,8,0.049644801020622256
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,4,0.0065760001540184024
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,4,0.029110398888587952
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,2,0.018694399297237395
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,1,0.00615679994225502
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,4096,1,0.012723200023174286
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,128,0.1658720016479492
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,3072,64,0.047337600588798524
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,64,0.08490239977836608
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,64,0.5487391948699951
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,128,1.1833888053894044
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,32,0.04830400049686432
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,16,0.024137599766254424
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,32,0.270579195022583
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,3072,8,0.03950079977512359
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,8,0.010524799674749374
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,4,0.008320000022649765
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,8,0.07041919827461243
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,4,0.039401599764823915
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,2,0.006377600133419037
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,6144,1,0.006374400109052658
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,1,0.016816000640392303
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,128,0.2165440082550049
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,128,1.6544031143188476
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,64,0.115283203125
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,4096,2,0.006364800035953522
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,32,0.06410239934921265
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,32,0.3600640058517456
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,64,0.7527999877929688
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,16,0.18383680582046508
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,8,0.012838399410247803
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,8,0.09473599791526795
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,4,0.008912000060081481
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,4,0.049439999461174014
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,2,0.006636799871921539
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,2,0.028880000114440918
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,1,0.006313599646091461
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,8192,1,0.01873600035905838
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,2,0.02301120012998581
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,128,2.5203680038452148
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,64,0.1539232015609741
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,8192,16,0.031097599864006044
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,64,1.1699968338012696
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,32,0.08865280151367187
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,6144,16,0.138153600692749
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,32,0.541974401473999
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,16,0.04346559941768646
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,8,0.023523199558258056
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,16,0.2713184118270874
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,128,0.32192320823669435
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,4,0.010604800283908844
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,8,0.13830080032348632
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,4,0.07090560197830201
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,2,0.03934400081634522
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,1,0.006707199662923813
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,12288,1,0.023030400276184082
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,128,0.43049278259277346
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,64,0.21919360160827636
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,64,1.6495679855346679
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,32,0.1171455979347229
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,128,3.4377696990966795
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,16,0.058169597387313844
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,32,0.7535871982574462
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,8,0.0301503986120224
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,8,0.1827455997467041
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,4,0.012556800246238708
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,4,0.09474560022354125
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,2,0.008790399879217148
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,2,0.04954879879951477
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,16384,1,0.006934399902820587
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,1,0.029212799668312073
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,12288,2,0.008646400272846222
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,64,0.2654560089111328
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,128,0.5332767963409424
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,32,0.1391968011856079
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,32,0.9653120040893555
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,16,0.07081599831581116
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,64,2.1071008682250976
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,16,0.45236802101135254
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,8,0.03734079897403717
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,4,0.016787199676036833
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,8,0.2275968074798584
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,2,0.008787199854850769
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,2,0.06000000238418579
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,float16,20480,1,0.008108799904584884
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,1,0.03324480056762695
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,16384,16,0.3600735902786255
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,4,0.11596800088882446
SGLang,0.5.9,NVIDIA GB300,mla_gen_post,default,fp8,20480,128,4.203331375122071
